Scrape Tokens

Scrape Tokens

Automate token market data collection from CoinGecko and CoinMarketCap using the agent-browser CLI.

Prerequisites

• agent-browser CLI installed and accessible
• Run agent-browser --help to verify installation

Core Workflow

Step 1: Scrape CoinGecko

Navigate to CoinGecko:

agent-browser open https://www.coingecko.com/

Take snapshot to identify elements:

agent-browser snapshot -i

CoinGecko provides:

• Top cryptocurrencies by market cap
• 24h price changes
• Trading volume
• Market cap rankings

Step 2: Extract Token Data (CoinGecko)

For the main token list, extract:

JavaScript extraction pattern:

const tokens = [];
document.querySelectorAll('table tbody tr').forEach(row => {
  const cells = row.querySelectorAll('td');
  if (cells.length > 5) {
    tokens.push({
      rank: cells[1]?.innerText,
      name: cells[2]?.querySelector('a')?.innerText,
      price: cells[4]?.innerText,
      change_24h: cells[5]?.innerText,
      market_cap: cells[8]?.innerText
    });
  }
});
return tokens;

Step 3: Scrape CoinMarketCap

Navigate to CoinMarketCap:

agent-browser open https://coinmarketcap.com/

Take snapshot:

agent-browser snapshot -i

CoinMarketCap provides:

• Similar data to CoinGecko
• Fear & Greed Index
• Trending tokens
• Recently added tokens

Step 4: Extract Token Data (CMC)

Extract similar fields from CoinMarketCap table.

Note: CMC may have different CSS selectors. Use snapshot to identify correct refs.

Step 5: Extract Trending & New Tokens

Both platforms show trending/new tokens. Capture these separately:

CoinGecko Trending:

agent-browser open https://www.coingecko.com/en/discover

CoinMarketCap Trending:

agent-browser open https://coinmarketcap.com/trending-cryptocurrencies/

Extract:

• Trending tokens (most searched/traded)
• Recently added tokens
• Top gainers/losers

Step 6: Compile Results

Merge and deduplicate data from both sources.

Output format (JSON):

{
  "scraped_at": "2025-01-26T12:00:00Z",
  "sources": ["CoinGecko", "CoinMarketCap"],
  "market_overview": {
    "total_market_cap": "$3.2T",
    "btc_dominance": "52.3%",
    "fear_greed_index": 72
  },
  "top_tokens": [
    {
      "rank": 1,
      "name": "Bitcoin",
      "symbol": "BTC",
      "price": "$104,500",
      "change_24h": "+2.5%",
      "change_7d": "-1.2%",
      "market_cap": "$2.1T",
      "volume_24h": "$45B"
    }
  ],
  "trending": [
    {
      "name": "Token Name",
      "symbol": "TKN",
      "price": "$0.45",
      "change_24h": "+150%"
    }
  ],
  "new_listings": [
    {
      "name": "New Token",
      "symbol": "NEW",
      "listed_date": "2025-01-25",
      "price": "$0.01"
    }
  ]
}

Step 7: Generate Report

Markdown format:

# Token Market Report - Jan 26, 2025

## Market Overview

| Metric | Value |
|--------|-------|
| Total Market Cap | $3.2T |
| BTC Dominance | 52.3% |
| Fear & Greed | 72 (Greed) |

## Top 10 Tokens

| Rank | Name | Price | 24h | 7d | Market Cap |
|------|------|-------|-----|-----|------------|
| 1 | Bitcoin (BTC) | $104,500 | +2.5% | -1.2% | $2.1T |
| 2 | Ethereum (ETH) | $3,200 | +1.8% | +3.5% | $380B |

## 🔥 Trending

1. **TokenA** (TKA) - $0.45 (+150%)
2. **TokenB** (TKB) - $1.20 (+85%)

## 🆕 New Listings

- **NewToken** (NEW) - Listed Jan 25 - $0.01

---
*Data from CoinGecko & CoinMarketCap*

Save files:

• tokens-YYYY-MM-DD.md
• tokens-YYYY-MM-DD.json

Step 8: Send Notification (Optional)

If telegram-notification MCP is configured:

Tool: mcp__telegram-notification__send_notification
Parameters:
  message: "📊 Market Report - Jan 26\n\n💰 BTC: $104,500 (+2.5%)\n💎 ETH: $3,200 (+1.8%)\n\n🔥 Trending: TokenA +150%"
  parse_mode: "Markdown"

Specific Token Lookup

To get data for a specific token:

# CoinGecko
agent-browser open https://www.coingecko.com/en/coins/bitcoin

# CoinMarketCap
agent-browser open https://coinmarketcap.com/currencies/bitcoin/

Extract detailed token info:

• Price history
• All-time high/low
• Circulating/total supply
• Contract addresses
• Exchange listings

agent-browser Quick Reference

Handling Pagination

For full token lists:

1. Identify pagination controls:

   agent-browser snapshot -i
   

2. Click next page:

   agent-browser click @e[next-button-ref]
   

3. Extract data from each page

4. Repeat until desired amount collected

Error Handling

• If price data unavailable, show "N/A"
• Handle rate limiting with delays between requests
• Retry failed requests once
• Log errors but continue with available data

Output Location

Save reports to current directory or specified path:

• Default: ./tokens-YYYY-MM-DD.{md,json}